A Gaussian approach (GA), which takes into account the influence of arbitrary optical and electrical filters, is proposed to analytically evaluate the performance of optically preamplified receivers. This GA also takes into account the effects of waveform distortion in the signal and in the signal-amplified spontaneous emission beat noise. Exact expressions for the mean and variance of the current at the decision circuit input are derived, which explicitly present the dependence on the optical and electrical filter transfer functions and take into account the presence or absence of a polariser. This GA is computationally much faster and simpler than more rigorous methods and the accuracy of its sensitivity estimates is investigated for different optical and electrical filters. It is shown that the GA predicts the optimal optical filter bandwidth very reasonably. Discrepancies below 1.2 dB are found between its sensitivity estimates and rigorous estimates. The assessment of the optical filter detuning impact on the receiver sensitivity is also investigated with this GA. Discrepancies less than 1 dB are found between the GA estimates and rigorous estimates. The GA sensitivity estimate's accuracy is also investigated in the presence of intersymbol interference (ISI) and very accurate predictions are obtained by the GA for significant ISI.

The outage probability (OP) in short-haul direct-detection (DD) optical links supported by weakly-coupled multicore fiber (MCF) impaired by intercore crosstalk (ICXT) and using ON-OFF keying signaling is assessed theoretically and through simulation. A closed-form expression for the OP is proposed for a single interfering core. This expression is useful to provide fast and reliable performance estimates in preliminary studies of weakly-coupled MCF-based systems where the ICXT is highly correlated along the signal bandwidth, i.e., when the skew-bit rate product is small. Particularly, the results for a 20-km-long link at 10 Gbit/s and walk-off below 1 ps/km are detailed. For high skew-bit rate products, i.e., when the ICXT is decorrelated along the signal bandwidth, the simulation results show that for a typical OP of 10 ^-4 , the tolerance to ICXT increases above 4 dB, when compared with low skew-bit rate products. It is also shown that the DD links with finite extinction ratio are more robust to outage induced by the ICXT than those with an ideal extinction ratio.

Intercore crosstalk (ICXT) impaired weakly-coupled multicore fiber (MCF) systems using on-off keying (OOK) and direct-detection (DD) are experimentally assessed. The stochastic properties of the received bit amplitudes that depend on the relative propagation delay between two cores (skew) of the MCF and on the OOK bit-rate are discussed. For low skew×bit-rate (≪1): (i) the received amplitude of bits 0 or 1 of the interfered core are mainly characterized by two possible levels; (ii) when the short-term average crosstalk (STAXT) power increases (decreases), the fluctuations of the received amplitudes of bits 1 and 0 become larger (lower). For high skew×bit-rate (≫1): (i) the received amplitudes of bits 0 and 1 present a noise-like behaviour; (ii) larger fluctuations of the received amplitudes of the OOK signal may not occur only for high STAXT powers. For both low and high skew×bit-rate, it is shown that the histogram of the received amplitudes of bits 1 or 0, and the histogram of the mean and variance of these amplitudes are reasonably well characterized by Gaussian distributions. Experimental results suggest that the rule of thumb to design emerging MCF applications with higher ICXT tolerance is to use cores with skew much longer than the symbol period.

The influence of the inter-core crosstalk (ICXT) on the performance of weakly-coupled multi-core fiber (WC-MCF) systems with coherent detection and arbitrary inter-core skew is studied numerically and analytically. We analyze the evolution of the instantaneous ICXT power, induced by polarization division multiplexed quadrature amplitude modulation signals, short term average crosstalk (STAXT), and detected ICXT components along the time to get insight on how the ICXT affects the coherent system performance. Numerical results show that, with low skew-symbol rate product (<>1), the STAXT is almost constant along the time and, although the detected ICXT varies along time, the detected ICXT variance is almost constant along time leading to very small fluctuations of BER along time. Analytical expressions for the variance of the coherently detected ICXT, average BER and optical signal-to-noise ratio (OSNR) penalty are proposed and shown to agree with Monte-Carlo simulation results for arbitrary skew, modulation format order and roll-off factor of the transmitted signals. Numerical results show that the maximum allowable ICXT level for 1 dB OSNR penalty increases when the skew augments and can reach 1.3 dB for a roll-off factor of 1. For most cases of interest of low roll-off factor, the increase of the maximum allowable ICXT level is very small. It is shown that the OSNR penalty estimates agree quite well with other authors' experimental results (with nearly zero roll-off factor).